View/Open

Linking Northern High-Latitude
Cryospheric Changes to Large-Scale
Atmospheric Circulation
Erlend Moster Knudsen
Dissertation for the degree of Philosophiae Doctor (PhD)
Geophysical Institute
University of Bergen, Norway
March 2015
Dissertation date: 0DUFK
Boredom didn’t frighten me. There are worse
pangs: the sorrow of not sharing with a loved
one the beauty of lived moments.
Solitude:
what others miss out on by not being with the
person who experiences it.
Sylvain Tesson,
The Consolations of the Forest
i
Acknowledgments
Numerous people have directly or indirectly helped me to get to the stage of writing my
PhD dissertation. Firstly, my main advisor Prof. Tore Furevik should get recognition for
balancing supervising me with travelling the world’s research institutions as the leader of
the Bjerknes Centre for Climate Research (BCCR) and the Centre for Climate Dynamics
(SKD). Secondly, my two other co-advisors, Dr. Yvan J. Orsolini at the Norwegian
Institute for Air Research and Prof. Nils Gunnar Kvamstø at the Geophysical Institute
(GFI), University of Bergen (UiB), also provided valuable contribution along the way. I
would like to especially draw attention to the former’s patience and constructive criticism
as significant in finalizing my work.
Thirdly, this work was financially supported by the Research Council of Norway
through the BlueArc project (no. 207650), and I wish to thank project leader Dr. Yongqi
Gao for a great topic and inspiring meetings.
Fourthly, both Prof. David W.J. Thompson and Prof. John E. Walsh deserve special
thanks. As hosts during my research stays at the Colorado State University (CSU), and
the International Arctic Research Center (IARC)/University of Alaska Fairbanks (UAF),
respectively, they both provided challenges and strengthened my ability to stand on my
own legs. In particular, I found Walsh’ ability to lead graduate students in finding good
solutions and becoming independent researchers being second to none.
Fifthly, I am highly grateful to my colleagues and fellow students at GFI/UiB, SKD/
BCCR, the Nansen Environmental and Remote Sensing Center, University of Oslo (during my shorter stays here), the Norwegian Research School in Climate Dynamics (over
short courses and seminars), CSU (during my 11 months research stay here) and UAF
(during my 2 months research stay here). In addition to coffee breaks and sports conversations, you have also had important contributions in sharing frustration and experiences
— dampening the feeling of solitude a PhD degree might give. Starting from scratch socially in both Fort Collins and Fairbanks, friends while at CSU and IARC/UAF deserve
extra thanks for making the transition from the Norwegian to the American culture as
smooth as possible.
Finally, family, friends and sport club buddies should get the last recognition. As
the blinders grew bigger and bigger towards the end of my PhD period, you gave me
understanding and room to focus.
Erlend Moster Knudsen
Bergen, December 2014
ii
iii
Abstract
Warming twice as fast as the rest of the globe, the northern high-latitudes arguably
show the clearest evidences of observed and projected climate changes. Two of these are
the rapid loss of Arctic sea ice and the shrinking of the Greenland Ice Sheet. In this
dissertation, three papers analyze the interaction between the cryospheric changes and
the atmospheric circulation.
In the first paper, summers of observed anomalous Arctic sea ice melt are composited in order to distinguish large-scale atmospheric patterns associated with sea ice loss.
While a positive cloud feedback characterizes warm high-latitude summers, storms track
more zonally in midlatitudes, leaving summers stormier, wetter and cooler in northwestern Europe and around the Sea of Okhotsk. Farther south, a heating band from the
Mediterranean Sea to East Asia indicates an increased probability of heat extremes in
these areas in summers of high Arctic sea ice melt.
The second paper analyzes projected changes in Arctic sea ice and Northern Hemisphere storminess. Here, a general poleward-shift of the main storm tracks is revealed.
Cyclone intensities generally follow cyclone frequencies, but with an elevated intensification over new open ocean areas. For most of the domain, precipitation increases significantly, with the highest enhancements along the poleward-shifted storm tracks and in the
sea ice diminishing Arctic.
Finally, the influence of cyclonic and anticyclonic activities on the observed Greenland
Ice Sheet (GrIS) surface mass balance (SMB) variability annually is analyzed in the third
paper. The synoptic features correlate with the SMB changes through their impact on
temperature and snow accumulation. Generally, enhanced cyclonic activity contributes
to increased snow accumulation over the GrIS by transporting more heat and moisture
from the south. A warming effect also results from enhanced anticyclonic activity in
summer, where fewer clouds increase the incoming shortwave radiation and thus surface
temperature. Overall, up to 60 % of the regional SMB variability can be explained by
these synoptic activities.
All together, the three papers illustrate how the Arctic sea ice and the GrIS interconnect with local and remote changes in the atmospheric circulation. While various studies
find the regional changes to significantly alter ecosystems, impact local communities and
enhance industry potential, distant consequences might include sea level rise, elevated
risks of extreme weather events, decline in agricultural production, economic disruption
and hence political instabilities.
iv
CONTENTS
v
Contents
Acknowledgments
Abstract
i
iii
1 Outline
1
2 Background
2
2.1
Sea ice cover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2
2.2
Sea ice thickness and volume . . . . . . . . . . . . . . . . . . . . . . . . . .
3
2.3
Seasonal variability of sea ice . . . . . . . . . . . . . . . . . . . . . . . . .
4
2.4
Arctic cyclonic activity . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6
2.5
Sea ice-atmosphere interaction . . . . . . . . . . . . . . . . . . . . . . . . .
7
2.6
Gaps in our understanding . . . . . . . . . . . . . . . . . . . . . . . . . . .
8
3 This study
3.1
3.2
9
Motivation and objectives . . . . . . . . . . . . . . . . . . . . . . . . . . .
9
Data sets and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9
3.2.1
Satellite data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.2.2
Reanalysis data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.2.3
Regional model data . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.2.4
Global model data . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2.5
Storm tracking algorithm . . . . . . . . . . . . . . . . . . . . . . . . 13
4 Summary of papers
14
4.1
Paper I: Observed anomalous atmospheric patterns in summers of unusual
Arctic sea ice melt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4.2
Paper II: Northern Hemisphere storminess in the Norwegian Earth System
Model (NorESM1-M) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
4.3
Paper III: The variability of surface mass balance over the Greenland Ice
Sheet and associated cyclonic and anticyclonic activities . . . . . . . . . . 16
5 Perspectives and outlook
17
5.1
Questions answered in this thesis . . . . . . . . . . . . . . . . . . . . . . . 17
5.2
Questions remaining unanswered . . . . . . . . . . . . . . . . . . . . . . . 18
5.3
The wider perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
vi
References
CONTENTS
24
Paper I: Observed anomalous atmospheric patterns in summers of unusual
Arctic sea ice melt
41
Paper II: Northern Hemisphere storminess in the Norwegian Earth System
Model (NorESM1-M)
91
Paper III: The variability of surface mass balance over the Greenland Ice
Sheet and associated cyclonic and anticyclonic activities
135
1
1
Outline
The thesis consists of an introductory part and a collection of three journal paper manuscripts.
A brief scientific background is given in section 2 before section 3 describes the study’s
objectives, tools and methods. The main results are given in section 4. Finally, section
5 sees the results in a wider perspective and offers an outlook for how future studies can
build on the results presented here.
The manuscripts included in the thesis are listed below.
Paper I
Observed anomalous atmospheric patterns in summers of unusual Arctic sea ice melt
Knudsen, E.M., Orsolini, Y.J., Furevik, T., Hodges, K.I.
Revised manuscript submitted to Journal of Geophysical Research
Paper II
Northern Hemisphere storminess in the Norwegian Earth System Model (NorESM1-M)1
Knudsen, E.M., Walsh, J.E.
Geoscientific Model Development Discussions, 7, 8975–9015
Paper III
The variability of surface mass balance over the Greenland Ice Sheet and associated
cyclonic and anticyclonic activities
Chen, L., Knudsen, E.M., Fettweis, X., Johannessen, O.M.
Manuscript in preparation
1
Reproduced with permission of the publisher
2
2 BACKGROUND
2
Background
The Arctic is home to arguably the most pronounced manifestation of climate change. It
is warming twice as fast as the global mean (Serreze et al., 2009; Bekryaev et al., 2010;
Screen and Simmonds, 2010). This is mainly due to strong radiative feedbacks such as tropospheric warming that deviates from the vertically uniform profile (lapse-rate feedback),
vertically uniform warming of the surface and troposphere (Planck feedback), replacement of white ice- or snow-covered surfaces by darker water, vegetation or soil (albedo
feedback), and to a lesser extent also dynamic feedbacks such as increased atmospheric
and oceanic poleward transports of heat and moisture (Perovich et al., 2007; Screen and
Simmonds, 2010; Riihel¨
a et al., 2013; Pistone et al., 2014; Pithan and Mauritsen, 2014;
Walsh, 2014).
2.1
Sea ice cover
A consequence of — and also a reason for — the anomalous warming in the Arctic is
the sea ice retreat (Screen and Simmonds, 2010; Screen et al., 2012; Cohen et al., 2014).
Since the start of the modern satellite era in the late 1970s, the mean sea ice extent at
the end of the melt season (September) has decreased substantially and at an accelerated
pace since the entry to the 21st century (Stroeve et al., 2012a). The present record low
of 3.4 · 106 km2 was set 13 September 2012, close to 50 % below the long-term mean
(1979-2000) (Liu et al., 2013; Parkinson and Comiso, 2013).
While most climate models project the summer Arctic sea ice to disappear by the
end of the current century (Figure 1) (Bo´e et al., 2009; Mahlstein and Knutti , 2012; Liu
et al., 2013; Stroeve et al., 2012b), some models reach this point already within two to
three decades (Wang and Overland , 2012). This state, referred to as the Blue Arctic, has
the Arctic Ocean covered by less than 1 · 106 km2 by September, with most of this ice in
the region north of Greenland and Ellesmere Island (Rothrock and Zhang, 2005; Maslanik
et al., 2007; Haas et al., 2008). However, the model spread and scientific discussion on the
timeline of a Blue Arctic is large (Bo´e et al., 2010; Liu et al., 2013). An ongoing debate
is whether a tipping point is reached (Smedsrud et al., 2008; Eisenman and Wettlaufer ,
2009; Maslanik et al., 2011; Tietsche et al., 2011; Wadhams, 2012). This “point of no
return” denotes a situation where sea ice formed in winter does not survive the summer
melt (Eisenman and Wettlaufer , 2009), even if the atmospheric forcing should return to
what was considered to be normal some decades ago.
None of the Coupled Model Intercomparison Project (CMIP) models are able to repro-
2.2 Sea ice thickness and volume
3
duce the observed trend in Arctic sea ice (Figure 2) (Stroeve et al., 2007; Wang and Overland , 2009; Stroeve et al., 2012b; Wang and Overland , 2012). Even though CMIP phase
5 (CMIP5) models are more consistent with observations than CMIP phase 3 (CMIP3)
models, they still underestimate the rapid decline (Stroeve et al., 2012b). Moreover, the
spread within CMIP5 models do not appear to be significantly reduced compared to that
of CMIP3 models (Stroeve et al., 2012b).
o
180 W
100
80
o
0
12
12 o
0W
90
E
70
60
50
40
o
60
60 o
E
30
W
20
Obs 1979−2005
Hist 1979−2005
RCP8.5 2037−2063
RCP8.5 2074−2100
0o
10
%
0
Figure 1: Map of observed and modeled Arctic sea ice concentration (in %) and boundaries
in September for three 27-year time periods. Shades of blue to white mark observed sea ice
concentration averaged over 1979–2005 from the National Snow and Ice Data Center (cf. section
3.2.1), with its boundaries in black. Red, blue and green lines show the corresponding modeled
sea ice boundaries from the Norwegian Earth System Model (cf. section 3.2.4) over the historical
1979–2005 and future scenario (following the RCP8.5 scenario; cf. section 4.2) 2037–2063 and
2074–2100 time periods, respectively. Boundaries are calculated using a threshold of 15 % sea
ice concentration.
2.2
Sea ice thickness and volume
The sea ice retreat is more dramatic than what the areal extent indicates. From 1991 to
2001, helicopter-borne electromagnetic-inductive measurements by Haas (2004) showed a
4
2 BACKGROUND
mean summer thinning of 23 % in the North Pole region, with an accelerated reduction
of up to 44 % by 2007 (Haas et al., 2008). This accentuates the rapid decline over the
last decades when compared to the Arctic-wide 1.6 m or 53 % summer thinning from
1958–1976 (submarine data) to 2003–2008 (satellite data) found by Kwok and Rothrock
(2009). Furthermore, using the Pan-Arctic Ice Ocean Modeling and Assimilation System
(PIOMAS), Overland and Wang (2013) estimated the September 2012 sea ice volume to
drop 72 % below the 1979–2012 mean.
This thinning and volume loss mainly follows from a regime shift from primarily multiand second-year ice (MYI and SYI; ice surviving previous summer melt) to first-year ice
(FYI; ice formed since last summer) (Maslanik et al., 2007; Haas et al., 2008; Kwok
et al., 2009; Maslanik et al., 2011). Compared to MYI and SYI, FYI is thinner, and
subsequently is more easily broken up and transported around — with a potential for
more rapid changes in sea ice cover (Maslanik et al., 2007; Haas et al., 2008). The more
mobile ice leaves converging or diverging transports with stronger impacts on air-sea fluxes
of momentum, heat and gas including moisture and greenhouse gasses (Thorndike and
Colony, 1982; Ogi et al., 2010; Spreen et al., 2011; Weiss, 2013). Hence, the thinning
might have more severe climatic consequences than the areal loss more easily detected by
satellites and ship traffic.
2.3
Seasonal variability of sea ice
Substantial intraannual and interannual variability characterize the Arctic sea ice. While
the sea ice extent traditionally has been in the range of 15–16 · 106 km2 of the polar
and subpolar oceans in March, it has been only 4–7 · 106 km2 in September (Parkinson
and Cavalieri , 2008; Cavalieri and Parkinson, 2012). In other words, the annual sea ice
minimum is less than half of its maximum extent. Nevertheless, the intraannual in the
Arctic is small compared to the Antarctic,, where only about 1/6 of winter sea ice remains
by the end of summer (National Snow and Ice Data Center , 2014a).
From an average value of about 7 · 106 km2 in the 1980’s and 1990’s, September
Arctic sea ice extent has dropped by about 13 % per decade (relative to the 1981 to 2010
average) to about 5 · 106 km2 over the last decade (Figure 2) (Cavalieri and Parkinson,
2012; National Snow and Ice Data Center , 2014b). Concurrently, March sea ice extent
has retreated by 3 % per decade (relative to the 1981 to 2010 average) (Vizcarra, 2014).
With the exception of the Bering Sea, the negative trend spans the whole Arctic, with
the highest yearly retreats found around the Sea of Okhotsk, Gulf of St. Lawrence and
the Kara and Barents seas (Cavalieri and Parkinson, 2012). Correspondingly, interannual
2.3 Seasonal variability of sea ice
5
spread is larger in months with the largest negative trend over the satellite period (Serreze
et al., 2007a; Cavalieri and Parkinson, 2012; Langehaug et al., 2013).
The fall is especially interesting from a climatological sense. At this time of year,
the atmosphere cools off rapidly, increasing the temperature gradient between ocean and
atmosphere due to the formers high thermal inertia. Because of the increasingly negative
trend in September sea ice cover and greater amounts of heat absorbed by the ocean
during the longer summer, it takes the sea ice on average more time each year to refreeze
(Francis et al., 2009a; Markus et al., 2009; Stroeve et al., 2014). Consequently, strong heat
and moisture fluxes from the warm ocean to the colder atmosphere are allowed in area
normally covered by sea ice, thus heating the lower troposphere and altering the static
stability (Francis et al., 2009a; Serreze and Barry, 2009; Strey et al., 2010; Cohen et al.,
2014).
Figure 2: Time series of observed (solid red line) and modeled (colored lines) Arctic sea ice
extent (in 106 km2 ) in September over 1900–2100. Dotted colored lines mark all 56 individual
ensemble members from 20 CMIP5 models, with individual model ensemble means in solid colored
lines. Solid black line denote the multi-model mean based on 38 ensemble members from 17
CMIP5 models, with ±1 standard deviation (STD) in dotted black lines. All model scenarios
follow the low emission scenario RCP4.5 (c.f. section 4.2). Inset shows observations compared
to multi-model means from CMIP3 and CMIP5 models, with ±1 STD in shaded color. From
Stroeve et al. (2012b).
6
2.4
2 BACKGROUND
Arctic cyclonic activity
A reduced static stability and increased contrasts between cold land and warm ocean
provide favorable conditions for cyclonic activity (Bengtsson et al., 2006). Cyclones are
found to be highly related to the Arctic sea ice (Sorteberg and Kvingedal , 2006), and they
have major impacts on precipitation, the radiation budget and cloudiness, in addition to
poleward heat and moisture transport (Bengtsson et al., 2006; Sorteberg and Walsh, 2008;
Budikova, 2009; Sepp and Jaagus, 2011).
While storm are stronger in wintertime (Bengtsson et al., 2009; Long and Perrie,
2012), summer is the most synoptically active period of the year over the central Arctic
Ocean (Serreze and Barrett, 2008), making summer cyclones crucial for high-latitude climate in this time of the year (Mesquita et al., 2008). This is exemplified by ‘The Great
Arctic Cyclone of August 2012’, which likely further amplified the record-breaking sea
ice minimum (Simmonds and Rudeva, 2012; Parkinson and Comiso, 2013; Zhang et al.,
2013). Earlier studies have shown that the cyclone tracks in summer are climatologically
more tightly defined and poleward located compared to storm tracks in winter (Mesquita
et al., 2008). In summer, the heating contrast between the Eurasian continent and the
Arctic Ocean helps to develop a strong cyclone activity over Northern Eurasia (Wernli
and Schwierz , 2006). This underlies a distinct summertime Arctic Ocean Cyclone Maximum (AOCM) (Serreze and Barrett, 2008; Orsolini and Sorteberg, 2009) that appears
in climatological storm track patterns. However, Screen et al. (2011) found this AOCM
to be missing in summers preceding Septembers of anomalously low sea ice area, in part
conflicting the previous results found by the same group (Simmonds and Keay, 2009).
Toward the end of the century, cyclones are generally expected to shift poleward (most
marked in the Southern Hemisphere) (Bengtsson et al., 2006, 2009; Lang and Waugh,
2011; Mizuta, 2012). No significant global changes in intensity are found, with the exception of a minor reduction in the number of weaker storms (Bengtsson et al., 2006,
2009). On the other hand, Bengtsson et al. (2009) reported an 11 % raise in cumulative
precipitation along the storm tracks, about twice the increase in the global average. The
overall precipitation enhancement is consistent with an increase of temperature and the
ability of warm air to contain more moisture (Knutti and Sedl´aˇcek , 2013), resulting in an
acceleration of the hydrologic cycle (Held and Soden, 2006).
Most studies of storm tracks under the changing climate have focused on midlatitudes.
Of the ones including high-latitudes, Bengtsson et al. (2006) and Orsolini and Sorteberg
(2009) examined the AOCM in future climates by applying Lagrangian cyclone tracking (cf. section 3.2.5) to the European Centre Hamburg Model (ECHAM) and Bergen
2.5 Sea ice-atmosphere interaction
7
Climate Model (BCM; cf. section 3.2.4), respectively. Bengtsson et al. (2006) reported
an intensification of summertime cyclones over the Arctic, but no significant changes in
frequency. A summertime intensification of cyclones was also found by Orsolini and Sorteberg (2009), but along with a raise in the number of cyclones, particularly over the
Siberian coast. They linked this feature to an enhanced meridional temperature gradient
between the Arctic Ocean and the warming Eurasian continent.
2.5
Sea ice-atmosphere interaction
Cyclones and external forcings (e.g., global warming) can significantly alter the Arctic
sea ice distribution (e.g., Bhatt et al., 2008; Deser and Teng, 2008; Francis et al., 2009b;
Balmaseda et al., 2010; Ogi et al., 2010; Long and Perrie, 2012; Overland et al., 2012;
Kapsch et al., 2013), but variability in the sea ice also feeds back into the other components
of the climate system (e.g., Francis et al., 2009b; Higgins and Cassano, 2009; Seierstad
and Bader , 2009; Kay et al., 2011; Vavrus et al., 2011; Liu et al., 2012b; Orsolini et al.,
2012; Sedl´aˇcek et al., 2012). Consequences for albedo and exchanges of heat, momentum
and water vapor are apparent (Budikova, 2009; Vihma, 2014, and references therein).
Changes in Arctic sea ice can be seen both locally and remotely. Local effects are primarily given as enhanced turbulent heat fluxes and outgoing longwave radiation (Serreze
et al., 2007b; Screen et al., 2014). Remote impacts are much more complex to disentangle
as they are often mixed with responses to many other local or remote factors that can
affect the climate in a given region at various space and time scales (Turner et al., 2007;
Strey et al., 2010; Screen et al., 2014). Hence, it is difficult based on observations alone
to isolate the effects of diminishing Arctic sea ice cover on midlatitude climate (Screen
and Simmonds, 2013; Cohen et al., 2014).
Nevertheless, Screen (2013) found anomalously low Arctic sea ice to be associated
with wet summers over northern Europe following southward shift of the jet stream in the
region. Moreover, Francis and Vavrus (2012) linked Arctic amplification to midlatitude
extreme weather through changes in the jet stream. Due to the polar warming, they found
a reduced poleward gradient in 1000–500 hPa thicknesses and an northward elongation
of ridge peaks in 500 hPa waves, resulting in more prolonged weather conditions favoring
stationary weather systems contributing to droughts or flooding, cold spells or heat waves.
Although other studies have found supporting results for the hypothesis (Tang et al.,
2013a; Coumou et al., 2014; Francis and Vavrus, 2014, in press), it is still heavily debated
within the scientific community (Barnes, 2013; Screen and Simmonds, 2013; Cohen et al.,
2014; Wallace et al., 2014; Walsh, 2014).
8
2 BACKGROUND
Questions also arise regarding the so-called warm Arctic/cold continents climate pattern. First introduced by Overland et al. (2011), the signal links Arctic sea ice melt to
continental snowfall and extreme winter weather. This pattern increases the frequency
of cold Arctic air outbreaks at lower latitudes with extreme low temperatures and often
snowstorms (Honda et al., 2009; Overland et al., 2011; Liu et al., 2012a; Outten and Esau,
2012; Song et al., 2012; Tang et al., 2013b; Guo et al., 2014). The weakening of the polar jet stream seems to be accompanied by a strengthening of the subtropical jet stream
(Cohen et al., 2013).
Various studies have indicated a relationship where anomalously low sea ice extent
in late summer increases the availability of atmospheric moisture and snowfall potential
(Cohen et al., 2012; Liu et al., 2012a). Building the Siberian snowpack in autumn would
strengthen and expand the Siberian High, with potential influence on the wintertime circulation mediated by the stratosphere (Wu et al., 2011). This results from the Eurasian
Siberian snow influence on the planetary Rossby wave vertical propagation from the troposphere to the stratosphere, a high snow anomaly leading to enhanced wave activity,
weakening of the polar vortex, and a lagged negative AO at the surface (e.g., Cohen
et al., 2013; Orsolini and Kvamstø, 2009). However, the complete link between sea ice
retreat, autumn snowfall and wintertime circulation anomalies is not fully understood.
2.6
Gaps in our understanding
Despite an increased focus on the changing Arctic sea ice from the scientific community,
media, politics and industry over the recent years, important questions remain unanswered. How much of the observed sea ice retreat can be explained by external forcing?
What role does internal variability play in the forcing and feedbacks of sea ice variability?
What causes the strong year-to-year fluctuations on top of the long-term negative trend?
How is the sea ice retreat affecting the cyclonic activity in the region? Is there a link
between the Arctic amplification and midlatitude extreme weather, and if so, what are
the physical mechanisms behind this?
It is not possible to answer all these questions in one PhD dissertation alone. Nevertheless, the quest for these answers provides the background and motivation for the
analyses carried out during the PhD period. While not providing final conclusions, this
PhD dissertation aims to significantly contribute to the discussions on the role of the
Arctic sea ice in the climate system.
9
3
This study
3.1
Motivation and objectives
The PhD dissertation is part of the research project Impact of Blue Arctic on climate
at high latitudes (abbreviated BlueArc), funded by the Research Council of Norway. Its
primary objective is to quantify the impact of the Blue Arctic on climate and climate
variability at high-latitudes. This will be achieved by comparing simulations of a future
blue Arctic Ocean high-latitude climate with the present-day situation to assess the impacts on the atmospheric and oceanic circulation, and to isolate and analyze anomalous
low sea ice events in observational and modeling data.
For reasons discussed in section 5.2, it is difficult to isolate the effects of retreating Arctic sea ice on atmospheric circulation in the papers constituting this thesis. Nevertheless,
the role of sea ice loss on climate variability is implicit in the results presented here, thus
contributing to the overall BlueArc project. In this thesis, satellite data, observationalbased reanalysis data and model data are used. The objectives of this thesis are:
• Isolate observed summers of anomalous Arctic sea ice extent to understand what
atmospheric mechanisms have caused and resulted from these anomalies.
• Increase the understanding of the interaction between synoptic atmospheric systems
and sea and land ice in the Arctic region.
• Evaluate changes in storm-related parameters in high-latitudes with projected climate changes.
The results of this study are expected to be of interest to a wide-ranging field. In
addition to contribute to the currently strong scientific interest of these topics, changes
in the cryosphere are of great importance to ecosystems, human infrastructure, industry
and politics. The Arctic is getting hot, for more than climatic reasons.
3.2
Data sets and methods
The Arctic is a remote region with relatively scarce in situ observations. Hence, the
analyses performed in this work heavily rely on data from satellites and model simulations.
10
3.2.1
3 THIS STUDY
Satellite data
Although ship-based records of sea ice extent in some regions extend several centuries back
in time (e.g., Vinje, 2001), even modern Arctic ship and buoy data are not sufficiently
comprehensive for measures of high-resolution as well as area-integrated variations of the
different sea ice quantities. Only satellites can provide the necessary information (Yang
et al., 2013, and references therein), although a circular mask over the geographic North
Pole is inevitable due to the orbit inclination and instrument swath (Cavalieri et al.,
2013).
Since 26 October 1978, satellites have provided information on nearly daily variations
in the Arctic sea ice cover. Lately (October 2010), the CryoSat satellite from the European
Space Agency (ESA) added sea ice thickness and hence volume to this picture.
In this dissertation, Arctic sea ice observations stem from the National Snow and
Ice Data Center (NSIDC) (Cavalieri et al., 2013). They are generated from brightness temperature data derived from five sensors. These are (in chronological order) the
Nimbus-7 Scanning Multichannel Microwave Radiometer (SMMR), the Defense Meteorological Satellite Program (DMSP) -F8, -F11 and -F13 Special Sensor Microwave/Imagers
(SSM/Is), and the DMSP-F17 Special Sensor Microwave Imager/Sounder (SSMIS). They
measure sea ice concentration (the fraction or percentage of ocean area covered by sea
ice; SIC), of which sea ice area (SIA) and extent (SIE) can be estimated using different
algorithms. While the SIA of a grid cell is the area multiplied by the SIC (threshold of
15 %, i.e., less ice is considered to be open water), the SIE calculation assumes each grid
cell is either “ice-covered” (SIC > 15 %) or “not ice-covered” (SIC ≤ 15 %). Hence, SIA
≤ SIE.
The NSIDC data are provided in the polar stereographic projection at a grid cell
size of 25 km x 25 km. While the SMMR sensor transmitted data every other day, the
subsequent sensors have provided data with daily resolution. For daily data analysis
in Paper I, missing data is therefore linearly interpolated from neighboring days. This
method was also used for the data gaps in August 1982, August 1984, December 1987
and January 1988.
3.2.2
Reanalysis data
All reanalysis data in this study stem from the European Re-Analysis Interim (ERAInterim) data set. The reason for this is the good and physically consistent representation of all the atmospheric fields when compared to the sparse observations in the regions
3.2 Data sets and methods
11
(Jakobson et al., 2012; Chung et al., 2013). Moreover, ERA-Interim is also found to be
among the best data sets for storm tracking (Hodges et al., 2011; Zappa et al., 2013),
although it captures fewer Arctic cyclones than the higher-resolution Arctic System Reanalysis (Tilinina et al., 2014).
ERA-Interim is produced by the European Centre for Medium-Range Weather Forecasts (ECMWF). The project was carried out in order to replace their previous atmospheric reanalysis, ERA-40, with improved methods to correct known problems of ERA40. ERA-Interim also includes improvements in number of pressure levels (from 23 to 37)
and added cloud parameters.
Described in more detail by Dee et al. (2011), ERA-Interim is a high-resolution reanalysis in space and time. Continuously updated in real time, it stretches from the start
of the modern satellite period in 1979, with 6-hourly, daily and monthly data available.
With T255 Gaussian grid (approximately 0.75◦ x 0.75◦ or 80 km x 80 km), it spans from
0.000◦ to 359.250◦ E and from −89.425◦ to 89.425◦ N. Its 60 model levels stretch from
the surface to 0.1 hPa. The analyses presented in this dissertation include data from
the surface, 850, 500 and 300 hPa levels, all interpolated to a 0.50◦ x 0.50◦ grid before
downloading.
The data assimilation system used to produce ERA-Interim is based on a 2006 release
of the Integrated Forecast System, cycle 31r2 (IFS-Cy31r2). This system includes a 4dimensional variational analysis (4D-Var) with a 12-hour analysis window.
3.2.3
Regional model data
In Paper III, the observationally constrained regional climate model Mod´ele Atmosph´erique
R´egional (MAR) is used for simulations of the Greenland Ice Sheet (GrIS) climate and
surface mass balance. Described in Gall´ee and Schayes (1994), the model is forced 6hourly by ERA-Interim at the lateral boundaries.